Most people who try your app will stop using it. That's true for every app, including the ones you love, and the number is higher than you'd guess.
Retention just means the fraction who keep coming back. It sounds like a marketing word and it's closer to the honest measure of whether your app is any good, because people don't return to things that don't help them.
Why it matters more than getting new users
If people leave as fast as you find them, you're running to stay still. Every new user costs effort, and if none of them stay, that effort resets to zero every month.
If people stay, each new user adds to a base that's still there next month. Growth stops being a treadmill and starts accumulating.
Which is why fixing retention before pushing hard on marketing is the right order. Spending time and money sending people into an app they'll abandon in a week is the most common way early effort gets wasted.
Reading the shape
If you look at retention properly, by group, as in the last lesson, you'll see one of three shapes.
It drops and then flattens out above zero. That's the good one. Some people leave, which is normal, but a core sticks around. That flat part is roughly what you have.
It slides steadily toward zero. Nobody's finding lasting value yet. More marketing won't fix it, it'll just move more people through the same leak.
Almost everybody leaves after one or two uses. This one usually isn't about your app's long-term value at all. It's the first minute, from lesson 3.2, which means people never got far enough to judge.
That third shape is the most common early on, and it's also the most fixable.
Most leaving happens immediately
The instinct is to think of churn as people getting bored after months.
Mostly it's people who never really started. They opened it, didn't understand it, didn't reach anything useful, and closed it. From your side that shows up as a user who signed up and vanished.
Which means retention work is usually onboarding work in disguise. Making month six better does nothing for someone who left on minute two.
The stages, and where to spend effort
People move through phases and each needs something different.
Newly arrived, they need to reach value fast. Actively using it, they need it to keep being worth it. Starting to drift, using it less, they're the ones worth catching, and it's much easier to keep somebody wobbling than to bring back somebody gone. Already gone, they're usually gone, though it's worth knowing why.
Most effort goes to the last group, sending win-back emails to people who've forgotten you exist. Better spent on the first and third.
Notifications, handled badly, make it worse
The reflex when people stop coming back is to remind them more.
Every notification either helps someone or spends a little of their patience. Helpful ones, that arrive when there's a real reason, keep working. Manufactured ones, sent because your numbers dipped, get you muted or uninstalled, and that permission never comes back.
The test before sending anything: is this for them, or for my chart? People can tell the difference more reliably than you'd expect.
The other kind of leaving
If you charge, some people leave without deciding to, because their card expired and the renewal failed silently.
This is a real slice of churn and it's the easiest to fix, since those people wanted to stay. Payment providers handle most of it if you turn it on: retry the charge, email the person, give them a way to update the card.
Worth checking that's switched on, because it's revenue you've already earned.
Ask the ones who go
When somebody cancels, one optional question about why is the cheapest research available.
Most won't answer. The ones who do tell you things no dashboard can, and it's often not what you expected. Frequently it's not the feature you were worried about. Sometimes it's price, sometimes it's one confusing thing they never got past.
A quick story. My app was designed around retention more carefully than anything else in it. The streak, the daily structure, the visible progress over ninety days, all of it existed to bring people back tomorrow.
I never found out if any of it worked, because it didn't launch, so there was never a cohort to follow.
The part worth taking: a retention design is a hypothesis until real people either come back or don't. You cannot test a habit on yourself, because you already care. The chart is the answer, and you only get the chart by shipping and waiting a few weeks.
If you want to work on retention
Point two decides everything else. The fix for "never got started" and the fix for "got bored in month three" have nothing in common.
Next: Keeping it going →
Go deeper: Chapter 50, Retention and Lifecycle